Closed janfreyberg closed 3 years ago
The paper is great, the only thing missing in my view is a clearer summary of the field - what options do people currently have? If there is no other open-source package or no other package that implements this workflow from start to finish, it would be worth saying so.
Review issue: https://github.com/openjournals/joss-reviews/issues/2474
Test completeness
I noticed that some tests check for existence of JPG output and some don't (e.g. here). It would be good to ensure this is tested everywhere a file gets deleted.
Documentation
There are some functions where I feel I don't understand the underlying algorithm / functionality from the documentation:
find_clumps
- no description is given of how a clump is identified. Is this done simply by "continuous" unbroken sets of damaged pixels? I wonder if this could be included.help()
on a function, I would prefer finding out the API and the conceptual functionality.plot_pixels_kfg
I feel I missed a section that explains K/F/G plots. This may be my relative ignorance, but I would appreciate a reference or explanation of how these plots are used. You describe this very well in the paper, would it be possible to copy small sections of the text to the vignette?detector
object attributes asslots
and demonstrates accessing them using integer indexing. I think it would be worth changing this to name access, as it's self-documenting.Code improvements
I note that the other review already mentions S3 objects and how detector objects are currently lists. It would be a nice improvement to make these actual S3 objects. This would allow for code improvements that would match up with how R users expect APIs to work. For example, this from the vignette:
could turn into this:
by the simple implementation of a summary method for detectors like so:
If
Detector
was an S3 object. I think this aligns better with R users, as callingsummary
and other base functions on objects is commonplace. The same could be done withplot_detector
->plot.Detector
This is a nit-pick and obviously involves considerable development for usability rather than functionality, though.